Silicon wafers are cut off from silicon single crystal ingots. It has been a recent trend to produce the silicon single crystal ingots by Czochralski method. FIG. 7 shows a front view of an unprocessed silicon ingot as it is after being produced by Czochralski method. The ingot is produced by the steps comprising: (1) dipping a seed crystal 10 (6 to 12 mm in diameter) in silicon melt; (2) pulling up the seed crystal to form a necking 12 (about 2 to 4 mm in diameter and about 50 to 100 mm in length) to eliminate lattice defects such as dislocations and processing defects; (3) forming a shoulder 14 to thicken the crystal to a diameter by about 1% larger than the desired diameter; (4) growing a straight cylindrical body 16 to a desired length; and (5) forming a conical tail 18.
The conical tail 18 is formed so that no dislocations due to thermal distortion are generated in the straight cylindrical body 16. In forming this conical tail 18, the temperature of the melt is raised a little and the pull-up speed is raised. Many dislocations appear in this conical tail 18 due to thermal distortion, often reaching to the straight cylindrical body 16.
A cylindrical useful crystal 20 is cut off from the ingot mentioned above and silicon wafers are produced by slicing this useful crystal 20. The cut-off step is that the shoulder 14 and the conical tail 18 are removed from the silicon ingot, followed by processing the periphery of the straight cylindrical body 16 to adjust its diameter to a desired value. Of the straight cylindrical body 16 (with a length of L1), the region 15 (with a length of L2) close to the shoulder 14 and the region 19 (with a length L4) close to the conical tail 18 are removed since dislocations may appear in these regions due to thermal distortion. Accordingly, the length L3 of the useful crystal 20 becomes shorter than the length L1 of the straight cylindrical body 16. While it is desirable to make the length L3 of the useful crystal 20 as long as possible for the purpose of effectively utilizing the expensive silicon ingot, the risk to incorporate the dislocation-appearing area into the useful crystal 20 is also enhanced when the removed regions 15 and 19 are diminished. Dislocations are liable to appear especially in the region 19 close to the conical tail 18. The regions 15 and 19 to be removed have been rather widely selected larger than the estimated dislocation areas which are based on the empirical estimation. This selection means that a significant amount of the good quality area including no dislocations had been discarded.
It would be possible to take out the maximum useful crystal containing no dislocations provided that the boundary between a dislocation-appearing area and a dislocation-disappearing area could be precisely found. Therefore, it is desired to provide a detection technique that can find the boundary.
There have been known in the art several methods for observing dislocations in a silicon single crystal: for example, (1) a wet-etching method applied to disc-shaped test pieces cut-off from the silicon ingot; and (2) an X-ray topography method applied to thin plates cut-off from the ingot, using forward-reflection X-ray topography (with a Lang camera). These methods, however, require the cut-off test pieces, so that dislocations could not be observed in the unprocessed silicon ingot as it is.
Dislocations could be observed without cutting off the silicon ingot by using, in principle, back-reflection X-ray topography. However, detection of dislocations by the back-reflection X-ray topography have been applied only for the disk-shaped test pieces cut-off from the silicon ingot and no X-ray topographic apparatus are known that can find the boundary between a dislocation-appearing area and a dislocation-disappearing area in the unprocessed silicon ingot which is used as a test sample as it is.
An object of the present invention is to provide a method and apparatus for X-ray topography that can observe dislocations and find the boundary between a dislocation-appearing area and a dislocation-disappearing area in the unprocessed single crystal ingot, typically silicon ingot, by using the ingot as it is as a test sample after being produced by Czochralski method.